TikTok Shop trend research via Sume trending search, $0.10 a call
Before a holiday product video, see what is trending for your product on TikTok. Sume's trending-videos search returns public video metadata at $0.10 a call.

To see what is trending on TikTok for a product before you make a holiday video, call POST /v1/trending-videos/search with a product keyword and a window such as this-week. Sume returns ranked public video metadata (watch URL, author handle, metrics, relevance), and each accepted call costs $0.10.
The timing matters this month: Best Buy says it debuts on TikTok Shop in late October 2026 with shoppable creator videos (read 2026-10-04), so the same keywords will get more crowded.
What does the endpoint return, and what does it not?
It returns metadata, not video files. The docs say it is a paid research utility rather than a generation model, and that it does not return downloadable source files. Do not plan to re-edit other people's videos with it; use it to read the hooks, formats and hashtags, then make your own.
| Field | Value |
|---|---|
| Platform | tiktok (the only value supported at launch) |
| query | Required in production; brand, product, creator or keyword, up to 200 characters |
| window | yesterday, this-week, this-month, last-3-months, last-6-months, all-time |
| limit | 1 to 50 in production, default 10 |
| Price | $0.10 per accepted call; summary_mode metadata included |
How do you run a search?
A search is one POST. Each entry carries a canonical watch URL, author handle, description, metrics and relevance; read the top ones and note the shared patterns.
import os, requests
r = requests.post(
"https://api.sume.com/v1/trending-videos/search",
headers={"Authorization": "Bearer " + os.environ["SUME_API_KEY"]},
json={
"query": "wireless earbuds",
"window": "this-week",
"region": "US",
"limit": 10,
"summary_mode": "metadata",
},
timeout=60,
)
r.raise_for_status()
print(r.text[:2000])
What do you do with the results?
Turn patterns into your own brief: how the first seconds are framed, how long the clips run, what text appears. Then generate an original clip with a model from the video docs, such as seedance-2.5 at 9:16. Keep claims about the product accurate; trend research shows format, not facts about your item.
How often should you search?
Once per product per planning pass is enough. Repeat browse calls may be served from cache, according to the docs, and every accepted call is billed, so batch your keywords rather than polling. Read the cost afterward with GET /v1/usage.
Sources
Related posts
More in Developers
- TikTok total_chunk_count rounds down: the final chunk can reach 128 MB
TikTok's Content Posting API sets total_chunk_count to video_size divided by chunk_size, rounded down, so the last chunk is bigger. Worked sizes and a checker.
- Track AI video spend per job: Synthesia Billing API vs Sume job cost
Synthesia added a Billing API and auto top-up. On Sume, every finished job reports usage.cost, so you can keep a per-job ledger without a billing endpoint.
- Transcribe a 30-minute file: OpenAI 25 MB cap vs Sume 10-minute jobs
A 30-minute file hits OpenAI's 25 MB upload limit and Sume's 10-minute duration hint. How to split once, transcribe in pieces, and re-base the word timings.
- Transcribe a 30-minute recording in three 600-second STT jobs: $0.33
Sume STT takes at most 600 seconds per job. Cut a 30-minute video into three 10-minute mp3 pieces with audio detach and transcribe each: $0.33 total.
Written by Sume